A Phenomenological Description of the Non-Fermi-Liquid Phase of MnSi
Extracted source text ยท arXiv cond-mat/0404732
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\documentstyle[aps]{revtex}
%----------------------------------------------------------------
\begin{document}
\draft
\title{
A Phenomenological Description
of the Non-Fermi-Liquid Phase of MnSi
\footnote{This Short Note will be published in J. Phys. Soc. Jpn.
Vol. 73 (2004) No. 11. }
}
\author{
O. Narikiyo
}
\address{
Department of Physics,
Kyushu University,
Fukuoka 810-8560,
Japan
}
\vskip 10pt
\date{
April 28, 2004
}
\vskip 10pt
\maketitle
%----------------------------------------------------------------
\begin{abstract}
In order to understand the non-Fermi-liquid behavior
of MnSi under pressure
we propose a scenario on the basis of the multispiral state
of the magnetic moment.
This state can describe the recent critical experiment
of the Bragg sphere in the neutron scattering
which is the key ingredient of the non-Fermi-liquid behavior.
\vskip 10pt
\noindent
Key Words: {\bf MnSi, non Fermi liquid, nesting,
helimagnet, multispiral, Bragg sphere}
\end{abstract}
%----------------------------------------------------------------
\vskip 30pt
MnSi is one of the most extensively studied materials
from the viewpoint of itinerant magnetism.
Recently its non-Fermi-liquid (NFL)
nature~\cite{Nature1,Nature2,Nature3}
which can not be understood
within the standard theory~\cite{Hertz,Moriya,Millis}
has been reported.
We will give a phenomenological description
of this NFL behavior in this Short Note
as a starting point for constructing a microscopic theory.
MnSi is a helimagnetic metal at ambient pressure
for the temperature $T$
below the transition temperature $T_{\rm c}=29.5$K.
$T_{\rm c}$ is suppressed
by the application of hydrostatic pressure $p$
and vanishes at the critical pressure $p_{\rm c}=14.6$kbar.
For $T>T_{\rm c}$ MnSi is a paramagnetic metal.
The phase transition changes its character
at $p=12{\rm kbar}\equiv p^*$.
For $p<p^*$ the transition is of second order,
while for $p>p^*$ it is of weakly first order.
It should be noted that a crossover line $T=T_0(p)$,
which merges to the transition line $T=T_{\rm c}(p)$ at $p=p^*$,
exists in $p-T$ plane.
For $T_{\rm c}<T<T_0$ the local magnetic moment exists
even in the paramagnetic phase.
Here $T_{\rm c}=0$ for $p>p_{\rm c}$.
These behaviors in $p-T$ plane are summarized,
for example, in Fig.~1 of ref.~3.
The NFL behavior in the DC resistivity~\cite{Nature1,Nature2}
is observed roughly for $T_{\rm c}<T<T_0$
as stated in the caption of Fig.~3 in ref.~3.
In this region
the resistivity $R$ is proportional to $T^{3/2}$.
For $p>p_{\rm c}$ where $T_{\rm c}=0$
this fractional power dependence persists down to a few mK.
This behavior is unexpected in two points.
Firstly
the transition at $T_{\rm c}$ is of first order for $p^*<p<p_{\rm c}$,
while such a fractional power dependence is expected
near the second order transition.
Secondly there is no characteristic scale of temperature
in the resistivity for $p>p_{\rm c}$,
while such a vanishing of the scale is expected
only at the critical pressure of the second order transition.
In order to understand the NFL behavior
a critical experiment by neutron scattering
has been reported very recently.~\cite{Nature3}
Across the first order phase transition line in $p-T$ plane
the unlocking of the spiral occurs.
Namely, it is observed that
the scattering intensity in the paramagnetic phase
spreads around the Bragg sphere
of radius $Q \sim 0.043$\AA$^{-1}$ in reciprocal lattice space,
while in the long-range ordered helimagnetic phase at ambient pressure
the magnetic Bragg peak is observed
at the point of $\vec Q = 0.037$\AA$^{-1}(1,1,1)$
which specifies the spiral of the magnetic moment in real space.
On the basis of this neutron scattering experiment
we try to explain the NFL behavior
in the temperature dependence of the resistivity
in a phenomenological manner.
Firstly we discuss the nature of the transition.
For $p<p^*$ the transition at $T=T_{\rm c}$
is a simple spin-density wave formation.
The local moment is absent for $T>T_{\rm c}$ in this pressure range
and the transition is of second order.
For $p>p^*$
non-vanishing global order parameter is attained only for $T<T_{\rm c}$,
while the local moment is formed for $T<T_0$.
In the region $T_{\rm c}<T<T_0$
a quasi-long-range correlation is developed,
although the global order parameter vanishes.
Since the local moment is already formed for $T>T_{\rm c}$,
the transition at $T=T_{\rm c}$ becomes first order in this pressure range.
The driving force for the global order is weak
and the transition is of weakly first order.
As an candidate for the driving force
we can think of the nesting of the Fermi surface,
since such a weak global effect
arises from the low-energy states around the Fermi energy.
Some nesting tendency of the Fermi surface
is seen in the band calculation.~\cite{band,JP}
It is naturally expected
that the preferred direction of the spiral
is determined by the nesting vector.
In the magnetically ordered state
the Fermi surface is gapped around the Bragg wave-vector.
The wave vector is determined by the nesting of the Fermi surface.
The gap is formed to gain the potential energy
in spite of the kinetic energy loss.
By the application of hydrostatic pressure
the relative importance of the kinetic energy increases
and the gap collapses at $T=T_{\rm c}$.
Across the first-order transition for $p>p^*$
the Bragg peak in the neutron scattering intensity
spreads over the Bragg sphere.
This unlocking of the spiral occurs
in order to gain kinetic energy.
The dissociation of $T_{\rm c}$ and $T_0$ for $p>p^*$
results from the fluctuation of the magnetic moments
due to the kinetic energy.
It should be noted that the absence of inversion symmetry in MnSi
related to B20 crystal structure is one of the most important features
of this material.
The helimagnetism is discussed to be derived
from the Dzyaloshinsky-Moriya interaction allowed for this symmetry
from the view point of localized magnetic moments.~\cite{NYHK,BJ}
On the other hand, MnSi is well described
as an itinerant magnet~\cite{Moriya,Fujimori}
so that the same effect is realized in the renormalized band structure
for quasiparticles.
Namely, the electronic state of MnSi is described as a Fermi liquid
with a quasiparticle band
into which the effect of localized magnetic moments
are renormalized.~\cite{Fujimori}
The renormalized quasiparticle band has nesting tendency.~\cite{band,JP}
Since we are interested in the low-temperature behavior
around the quantum critical point,
the approach on the basis of the Fermi liquid is more appropriate
than that on the basis of the localized magnetic moments.
Since a formation of the Bragg circle
has been theoretically reported in a two-dimensional case
on the basis of a double spiral state~\cite{double1,double2}
where the spiral of the magnetic moment is specified by two basis vectors,
we expect that the Bragg sphere is derived
from some multispiral state.
The microscopic derivation of the multispiral state
needs detailed information on the Fermi surface
and should be done in a future study.
The absence of the characteristic temperature
even in the region away from the quantum critical point
can be understood by the mechanism of the gradual change of the wave vector
specifying the multispiral state,
while the nesting vector specifying the Bragg peak is uniquely defined.
Thus in the wide range for $p>p_{\rm c}$
the coherence length becomes divergently large on the Bragg sphere
in the limit of $T \rightarrow 0$.
In this paper we use the word, the coherence length, after ref.~3
and it should be distinguished from the true correlation length
as discussed below.
The microscopic derivation of the above mechanism
on the basis of the gradual change of the Fermi-surface shape
should be also done in a future study.
Next we discuss the NFL nature
observed in the temperature dependence of the resistivity.
In the wide region of interest, $p>p_{\rm c}$,
the coherence length can be assumed to be divergently large~\cite{Nature3}
in the limit of $T \rightarrow 0$.
Thus the coherence length $\xi$ is approximately modeled as
%============================================
\begin{equation}
\xi^{-2} \sim \alpha T,
\label{xi}
\end{equation}
%============================================
where the right-hand-side term is induced
by the mode-mode coupling~\cite{Moriya,Millis}
and $\alpha$ is a temperature-independent constant.
This form is derived for non-critical state,
while we have approximately set
$\xi^{-2} \rightarrow 0$ when $T \rightarrow 0$
in consistent with the resolution-limited scattering intensity.~\cite{Nature3}
With this $\xi$ the imaginary part of
the dynamical spin susceptibility at low energy
can be modeled as
%============================================
\begin{equation}
\chi''({\vec q},\omega)
= { D \omega / \Gamma \over
[ \xi^{-2} + A({\vec q}-{\vec Q})^2 ]^2 + (\omega / \Gamma)^2 },
\label{chi}
\end{equation}
%============================================
where ${\vec Q}=Q(1,1,0)$ with $Q \sim 0.043$\AA$^{-1}$.
$A$ and $D$ are constants
independent of ${\vec q}$, $\omega$ and $T$.
Here we have taken into account the fact
that the notion of the Bragg sphere is an approximate one and
that the scattering intensity is strongest at this ${\vec Q}$
as observed in the experiment.~\cite{Nature3}
The parameter for the dissipation $\Gamma$ is
of the order of $v_{\rm F}Q$
where $v_{\rm F}$ is the renormalized Fermi velocity for quasiparticles.
This form is valid at low energy of $\omega < \Gamma$
and sufficient for the discussion
of the low-temperature behavior of the resistivity.
It should be noted that the coherence length $\xi$
used in this paper is effective only for the modeling
of the imaginary part of the susceptibility
at low energy of $\omega < \Gamma$.
Actually the divergent behavior
in the neutron scattering intensity~\cite{Nature3}
is related to the imaginary part of the susceptibility
in the low-energy limit.
The true correlation length or the thermodynamic susceptibility
is determined by the real part, $\chi(\vec q)$ at $\omega = 0$,
which is related to the imaginary part by the Kramers-Kronig relation as
$ \chi(\vec q) = {1 \over \pi} \int_{-\infty}^\infty {\rm d}\omega
\chi''({\vec q},\omega) / \omega $.
The behavior of the true correlation length is determined
not only by low-energy process but also by high-energy process
and does not necessarily coincide with the coherence length
characterizing the low-energy behavior.
Such a situation is similar
to the case of cuprate superconductors.~\cite{NM}
Using this $\chi''({\vec q},\omega)$
of $z=2$ form, where $z$ is the dynamical exponent,
we calculate the resistivity $R$.
It is easily evaluated for our non-critical state
as a correction to the Fermi-liquid behavior~\cite{KY} as
%============================================
\begin{equation}
R \propto T^2 \sum_{\vec q}
{ 1 \over [ \xi^{-2} + A({\vec q}-{\vec Q})^2 ]^2 }.
\label{R}
\end{equation}
%============================================
Here the extracted factor $T^2$ represents the Fermi-liquid behavior
and is consisitent with the form of the dissipative term $\omega/\Gamma$
in $\chi''({\vec q},\omega)$.
The summation in the right-hand side representing the correction
is proportional to the integral
%============================================
\begin{equation}
\int_0^{p_{\rm c}}
{ p^2{\rm d}p \over ( p^2 + \xi^{-2} )^2 }
\sim \xi
\sim (\alpha T)^{-1/2},
\label{int}
\end{equation}
%============================================
where we have set as ${\vec p}={\sqrt A}({\vec q}-{\vec Q})$.
Thus we obtain
%============================================
\begin{equation}
R \propto T^{3/2},
\label{RT}
\end{equation}
%============================================
in the low-temperature limit.
This form of resistivity scaling is valid
for $T < \Gamma$
where the assumed form of $\chi''({\vec q},\omega)$ is dominant
in the evaluation of the resistivity.
Since $v_{\rm F}$ and $Q$ depend only weakly on the applied pressure,
the crossover temperature ${\tilde T} \equiv \Gamma$
also depends only weakly on the pressure
in consistent with the experiment.~\cite{Nature1,Nature2,Nature3}
In this Short Note
we have proposed the multispiral state
in order to explain the key experiment of
the Bragg sphere in the neutron scattering.
A microscopic description of such a state
needs detailed information on the Fermi surface
and is left as a future study.
On the basis of the quasi-long-range correlation
observed as the Bragg sphere
the NFL behavior in the temperature dependence
of the resistivity is naturally understood.
\vskip 30pt
%----------------------------------------------------------------
\begin{references}
\bibitem{Nature1}
C. Pfleiderer, S. R. Julian and G. G. Lonzarich:
Nature \textbf{414} (2001) 427.
\bibitem{Nature2}
N. Doiron-Leyraud, I. R. Walker, L. Taillefer, M. J. Steiner,
S. R. Julian and G. G. Lonzarich:
Nature \textbf{425} (2003) 595.
\bibitem{Nature3}
C. Pfleiderer, D. Reznik, L. Pintschovious, H. v. L\"ohneysen,
M. Garst and A. Rosch:
Nature \textbf{427} (2004) 227.
\bibitem{Hertz}
J. A. Hertz:
Phys. Rev. B \textbf{14} (1976) 1165.
\bibitem{Moriya}
T. Moriya:
\textit{Spin Fluctuations in Itinerant Electron Magnetism}
(Springer, Berlin, 1985).
\bibitem{Millis}
A. J. Millis:
Phys. Rev. B \textbf{48} (1993) 7183.
\bibitem{band}
L. Taillefer, G. G. Lonzarich and P. Strange:
J. Mag. Mag. Mat. \textbf{54-57} (1986) 957.
\bibitem{JP}
T. Jeong and W. E. Pickett:
cond-mat/0403442.
\bibitem{NYHK}
O. Nakanishi, A. Yanase, A. Hasagawa and M. Kataoaka:
Solid State Commun. \textbf{35} (1980) 995.
\bibitem{BJ}
P. Bak and M. H. Jensen:
J. Phys. C \textbf{13} (1980) L881.
\bibitem{Fujimori}
J-Y. Son, K. Okazaki, T. Mizokawa, A. Fujimori,
T. Kanomata and R. Note:
J. Phys. Soc. Jpn. \textbf{71} (2002) 1728.
\bibitem{double1}
C. L. Kane, P. A. Lee, T. K. Ng, B. Chakraborty and N. Read:
Phys. Rev. B \textbf{41} (1990) 2653.
\bibitem{double2}
B. Chakraborty, N. Read, C. L. Kane and P. A. Lee:
Phys. Rev. B \textbf{42} (1990) 4819.
\bibitem{NM}
O. Narikiyo and K. Miyake:
J. Phys. Soc. Jpn. \textbf{63} (1994) 4169.
\bibitem{KY}
H. Kohno and K. Yamada:
Prog. Theor. Phys. \textbf{85} (1991) 13.
\end{references}
%----------------------------------------------------------------
\end{document}
%----------------------------------------------------------------
\bye
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%